Re: Portability issues in shm_mq

I wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
>> First, can you retest this with the latest code?

> Yeah, on it now.

Early returns not good:

*** /Users/buildfarm/bf-data/HEAD/pgsql.93630/contrib/test_shm_mq/expected/test_shm_mq.out      Tue Mar 18 12:00:18
2014
--- /Users/buildfarm/bf-data/HEAD/pgsql.93630/contrib/test_shm_mq/results/test_shm_mq.out       Tue Mar 18 12:17:04
2014
***************
*** 5,18 **** -- internal sanity tests fail. -- SELECT test_shm_mq(32768, (select
string_agg(chr(32+(random()*96)::int),'') from generate_series(1,400)), 10000, 1);
 
!  test_shm_mq 
! -------------
!  
! (1 row)
!  SELECT test_shm_mq_pipelined(16384, (select string_agg(chr(32+(random()*96)::int), '') from
generate_series(1,270000)),200, 3);
 
!  test_shm_mq_pipelined 
! -----------------------
!  
! (1 row)
! 
--- 5,12 ---- -- internal sanity tests fail. -- SELECT test_shm_mq(32768, (select
string_agg(chr(32+(random()*96)::int),'') from generate_series(1,400)), 10000, 1);
 
! ERROR:  message corrupted
! DETAIL:  The original message was 400 bytes but the final message is 7492059346764176 bytes. SELECT
test_shm_mq_pipelined(16384,(select string_agg(chr(32+(random()*96)::int), '') from generate_series(1,270000)), 200,
3);
! ERROR:  message corrupted
! DETAIL:  The original message was 270000 bytes but the final message is 7492059347033776 bytes.


This is C locale on a 32-bit machine, so you'll likely be seeing the same
complaint in already-online buildfarm members.

Note that size_t is definitely not int64 on this machine, so it looks to
me like your int64-ectomy was incomplete.  Those message lengths should
be impossible no matter what on this hardware.
        regards, tom lane